Research map / Image generation / Image restoration
DWDN: Deep Wiener Deconvolution Network for Non-Blind Image Deblurring
Jiangxin Dong, Stefan Roth, Bernt Schiele
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add feature-based Wiener deconvolution to the network (ablation) | DWDN · Levin 1% noise | PSNR: 25.70 -> 36.90 | Improved |
| add gradient features to intensity in classical Wiener deconvolution (ablation) | Wiener_I · Levin 1% noise | PSNR: 30.32 -> 33.37 | Improved |
| use 9 learned deep features instead of intensity plus gradients (ablation) | Wiener_I+G · Levin 1% noise | PSNR: 33.37 -> 34.09 | Improved |
| learn 16 deep features instead of 9 (ablation) | Wiener_9 features · Levin 1% noise | PSNR: 34.09 -> 34.78 | Improved |
| add multi-scale feature refinement on top of Wiener with deep features (ablation) | Wiener_16 features · Levin 1% noise | PSNR: 34.78 -> 36.90 | Improved |
| use learned deep features instead of input image + gradients (ablation) | DWDN_I+G · Levin 1% noise | PSNR: 36.52 -> 36.90 | Improved |
| use piecewise linear (ReLU) deep features instead of linear features (ablation) | Wiener_16 features · Levin 1% noise | PSNR: 33.69 -> 34.78 | Improved |
| add multi-scale refinement (ablation) | DWDN · Levin kernels 53-59, 1% noise | PSNR: 27.45 -> 27.73 | Improved |
| add 2 cascades (ablation) | DWDN · Sun et al. 1% noise | PSNR: 34.05 -> 34.27 | Improved |
| add 3 cascades (ablation) | DWDN · Sun et al. 1% noise | PSNR: 34.05 -> 34.47 | Improved |
| add residual learning in earlier cascade stages (ablation) | DWDN+ · Sun et al. 1% noise | PSNR: 34.47 -> 34.62 | Improved |
| train one noise-blind model instead of noise-specific models (ablation) | DWDN · Levin 1% noise | PSNR: 37.02 -> 36.90 | No clear effect |
Later papers that compared with this method
3 later accepted papers report DWDN as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- A Physics-informed Low-rank Deep Neural Network for Blind and Universal Lens Aberration Correction (CVPR 2024 · better on 0 of 2 shared results)
- Deep Richardson–Lucy Deconvolution for Low-Light Image Deblurring (IJCV 2023 · better on 5 of 5 shared results)
- Learning Deep Non-Blind Image Deconvolution without Ground Truths (ECCV 2022 · better on 1 of 3 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Deep Wiener Deconvolution: Wiener Meets Deep Learning for Image Deblurring (NeurIPS 2020 · similarity 0.96)
- Deep Image Deblurring: A Survey (IJCV 2022 · similarity 0.90)
- Neural Blind Deconvolution Using Deep Priors (CVPR 2020 · similarity 0.89)
- Learning Fully Convolutional Networks for Iterative Non-Blind Deconvolution (CVPR 2017 · similarity 0.89)
- Learning Discriminative Shrinkage Deep Networks for Image Deconvolution (ECCV 2022 · similarity 0.89)
- DeblurDiff: Real-Word Image Deblurring with Generative Diffusion Models (NeurIPS 2025 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
